We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Classification of recycled plastics using sparse and imbalanced spectral data and data augmentation by the generative adversarial network
AI summary Read the abstract
Using GAN-generated synthetic FTIR spectra to balance training datasets, researchers achieved 96.2% classification accuracy for six recycled polymer types, substantially outperforming models trained on imbalanced data alone. Accurate rapid identification of plastic types is essential for improving recycling rates and reducing the quantity of misidentified plastics that end up fragmenting into environmental microplastics.
Accurate identification of post-consumer plastics is essential to establishing high-performance recycling processes and enabling a circular and sustainable economy and environment through effective recycling and remanufacturing. However, Fourier transform infrared (FTIR) spectra of recycled materials often exhibit noise, baseline shifts, and overlapping signatures from additives or contaminants, resulting in datasets that are both sparse and severely imbalanced. This data complexity, sparsity, and class imbalance can degrade conventional machine-learning classifiers, resulting in higher rates of misclassifying plastics. To address these challenges, we investigated if data augmentation using generative adversarial networks could enhance polymer classification performance. We implemented a Generative Adversarial Network (GAN) framework that integrates adversarial training with a classifier-guided feedback loop to synthesize realistic, class-discriminative FTIR spectra for six commonly recycled polymers, polyethylene (PE), polypropylene (PP), polystyrene (PS), polycarbonate (PC), polyethylene terephthalate (PET), and acrylonitrile butadiene styrene (ABS), and trained multilayer perceptron classifiers on datasets with varying ratios of synthetic data. The optimal balanced accuracy of 96.2% was achieved when synthetic spectra accounted for 50% of the training set, whereas including more than 90% synthetic data degraded generalization. Synthetic data augmentation using a GAN with the optimal augmentation ratio improved ABS classification accuracy, precision, and recall by 43%, 50%, and 33%, respectively, compared with no augmentation and replicate experimental measurements. These results demonstrate that GAN-based data augmentation can effectively mitigate data sparsity and class imbalance in spectral classification of common plastics, providing a practical foundation for creating robust online polymer classification systems.
More Papers Like This
FRDA: Fingerprint Region based Data Augmentation using explainable AI for FTIR based microplastics classification
AI summary Read the abstract
Researchers developed a new machine learning approach called FRDA that uses explainable artificial intelligence to improve the accuracy of microplastic identification from infrared spectroscopy data. The method generates realistic synthetic training data focused on the most chemically informative regions of the spectra, helping overcome the problem of limited and imbalanced datasets. The technique significantly improved classification accuracy, especially for difficult-to-identify plastic copolymers and mixtures.
GANsemble for Small and Imbalanced Data Sets: A Baseline for Synthetic Microplastics Data
AI summary Read the abstract
Researchers developed a generative adversarial network (GAN) ensemble approach to generate synthetic training data for microplastic image classification, addressing the challenge that real microplastic image datasets are small and imbalanced by polymer type. The synthetic augmentation improved classifier accuracy and recall, particularly for underrepresented plastic categories.
GANsemble for Small and Imbalanced Data Sets : A Baseline for Synthetic Microplastics Data
AI summary Read the abstract
Researchers applied a generative adversarial network ensemble approach to address the problem of small and imbalanced microplastic datasets in machine learning, producing synthetic data to improve model training for plastic particle classification.
Generation of synthetic FTIR spectra to facilitate chemical identification of microplastics
AI summary Read the abstract
Researchers generated synthetic FTIR spectra of microplastics using computational methods to augment training datasets for automated spectral identification algorithms. The synthetic spectra closely matched experimentally measured spectra, and classifiers trained on augmented datasets showed improved accuracy for identifying underrepresented polymer types in real-world samples.
Deep learning for chemometric analysis of plastic spectral data from infrared and Raman databases
AI summary Read the abstract
A novel deep learning architecture called PolymerSpectraDecisionNet was trained to identify common recyclable plastics from infrared and Raman spectral databases. The model outperformed conventional chemometric methods for polymer classification and was designed to handle real-world spectral variability relevant to the plastics recycling industry.
Research digests by email
When a large batch of papers lands in the Atlas, we read through it and send a short write-up of what stood out.